I remember as a kid, not being able to afford anything more than the time to recycle bits from dead electronics, I'd sit for hours hand matching sorting resistors by actual value for when I'd need an exact value... ugh........ today's resources are amazing.
Dear Dave, as already mentioned below you should play with the gate resistor. Also you should add a GateDrain (Not Gate Source!) capacitor in a small value. (47p-100p). If you want to do a PCB: Also you already you want to use a excellent opamp you should consider using the reference as reference (datasheet: typical application) and the opamp not only as a impedance converter but as regulator/amplifier. You already use a excellent part, so use the incredible Gain-Bandwith product you have provided by that part. But be aware that the offset for your opamp increases when you have to high rail voltages. (Refer datasheet) One more thing to tweak precision, but handle these advises with care: For the ground connection of the power paths you should do a star-grounding, NOT returning over the sense pin. The sense pin is not meant for current. Instead I would recommend to use your spare OpAMP as impedance converter for the negative sense terminal and drive the reference ground. Greetings from Germany.
Great vid! The "how to make a circuit that does XYZ" type videos are always appreciated by those of us with a software / digital electronics background! -MCM
A compensation network is needed when putting a pass element like mosfet here. This is because it adds a phase lag. Of course you can just throw random caps to attenuate high frequencies and hope it will work well. But it does not guarantee stability. Only freq. response analysis over a range of temperature and loads (and supply voltage!) can ensure stability problem is solved. Also, gate resistance looks way bigger than it normally is. Most schemes use 10-100 Ohm gate resistors. Larger values may not actually solve problem, it hides it. This scheme is a perfect example of it: 12k was not enough, it needs a cap in the right place. Another thing is that electrolytic caps are bad in compensation loops as they have poor frequency response, bad tolerance and dry out.
This is awesome, I watch this blog all the time and just out of pure coincidence I was about to looking for a precision 1amp current source when I opened up my laptop earlier today!!
bravo, Dave I've always shied away from testing precision circuits on a plug-board like that. Thank you for showing that with just a little care, one can get meaningful results and some confidence that the properly-spun PCB WILL meet spec.
When I first started watching I didn't get anything but now I understand maybe... 20% or so. It's a strange way of learning, but hey.. it's free and enjoyable :P
The 12k gate resistor and the input capacitance of the mosfet creates a pole in the transfer function. The phase shift associated with the pole is causing the oscillation. The solution is to decrease the value of the gate resistor. This will move the pole much higher in frequency.
the solderless breadboard you are using is "riddled" with inductance and stray capacitance. You need to hard-wire the circuit and keep the circuit tight.
What was the frequency range you were getting in the noise (couldn't see the scope time axis value) ? To start finding a value, I typically would then choose the RC time constant in something in that order of magnitude times some factor (for 470µF and 1.25Ω , RC is ~0.6ms).
that's what I would have done as well, would like it if he gets back to us about the switching residual instead of playing Russian roulette with those capacitors :)
EEVblog if you short the V+ out in the field there isn't anything else to stop it from dump all the current the supply has on hand. Might be light for a battery powered thing like yours, mine is running off a filtered ATX power supply. Believe it can do 15A on the 12V rail? Yeah, $1 peace of mind! :-D
I'd like to build this but can't find a source for a 4 wire 1.25 ohm resistor. how much would I be messing this up if I just tapped sense off the pins of a 2 wire resistor? (ima noob)
At 25:00 Why shows the amp-meter 1.112A while the power supply's current limit kicks in at 1.100 A? Which is even visible on the PS. Both devices were pretty close in value before and after this test. What's going on here?
The 2N2222 have an absolute rating of 800mA continuous collector current (based on farnell datasheet). You want to get 1A out of it That's probably why it oscillates.
And in no meaning You gave away for what You need that current source! ;-) I guess 1 amp on four D-Cells: You're going to build a torch or an electrical foot warmer. :-)
I don't understand why Dave didn't just use one op amp (low offset voltage) driving one N channel MOSFET. Non inverting input to the op amp connected to the precision Vref. The 4 terminal R on the source of the MOSFET to Gnd. Why use that feedback loop in the chip? Can't draw a circuit diagram here but I think there's a much simpler circuit that would do the same job.
Would increasing the load inductance help with stability ? Is it safe to completely short the load terminals ? How could you implement overvoltage protection at the load terminals (say someone sticks a charged capacitor on there) How does it respond to high load resistance ? What is the min-max load resistance where the current will be in spec ? Can it be damaged with ESD ?
So, when you changed the 10uf output cap to a 470uf cap, you matched the haromics of the suggested 4.7uf cap? Maybe you are just in the wrong octave O:-)
If the goal is to avoid that 2mA sink into the sense out, why not just use the path from the bottom of the load to ground as output, rather than the top? Would that allow you to avoid the extra opamp? Besides not giving you the added stability, of course.
Hi. (I'm korean so my English is not good ^^;;) I make a variable current source. It makes current ~1A. But there's a problem that is too hot from the mosfet. It's using 24V source for 0~20ohm Load. How about you??
+Thomas YT A 4 pin or 4 wire "Kelvin" connection is a more precise method of measuring to reduce error that could come into play due to resistances along the main current path. Search for "4 wire kelvin" on youtube. Afrotechmods and TechnologyCatalyst have very good tutorials on it to get you started
+Mike Novo Thanks, Mike. I think, I understand the principle. In the case of a 4 pin resistor, are two pins simply interconnected, so that I have one pair of contacts for the constant current source and one pair for measuring the voltage?
Exactly. The two inner pins are normally use to measure voltage and are trimmed for accuracy. If you don't have a 4 pin Shunt resistor, You want to solder your sense leads as close as possible to reduce error. I am very new at this and I'm sure others can chime in with more pearls of wisdom. Enjoy
Glad you found a work-around. However, what of the original design provided by Linear? I very much doubt they'd publish something that didn't work. "Not even half a bee's dick" is a new one on me. We have another here in the US: "RCH" (red ---- hair). Quite vulgar, really. Among the crass, hormone-driven, insecure, semi-pubescent men in the electronics field encountered when I was growing up, this unit of measure came to frequent use.
Nice Video - but. 24;30 - Do you know what "good enough" means? And please! Never ever ever do a Dave CAD with this black pen - it just don't look genuine ;)
Oh my god. Dave I like your videos, but well your circuit is again so bad. Anyway you connected at least pin 8 to ground now. Your circuit is for high side current source and you connect DMM on top of that ... and fixing that bug with an additional OPV is very bad. And there is more. You are feeding your complete GND current through the sense wire. You are wasting all those precision components by bad design.
I remember as a kid, not being able to afford anything more than the time to recycle bits from dead electronics, I'd sit for hours hand matching sorting resistors by actual value for when I'd need an exact value... ugh........ today's resources are amazing.
As a beginner electric engineer I found this video very informational.
thanks Dave. I learn so much from the way you walk through, step by step, your thought process on the bread board.
Dear Dave, as already mentioned below you should play with the gate resistor. Also you should add a GateDrain (Not Gate Source!) capacitor in a small value. (47p-100p).
If you want to do a PCB:
Also you already you want to use a excellent opamp you should consider using the reference as reference (datasheet: typical application) and the opamp not only as a impedance converter but as regulator/amplifier. You already use a excellent part, so use the incredible Gain-Bandwith product you have provided by that part. But be aware that the offset for your opamp increases when you have to high rail voltages. (Refer datasheet)
One more thing to tweak precision, but handle these advises with care:
For the ground connection of the power paths you should do a star-grounding, NOT returning over the sense pin. The sense pin is not meant for current.
Instead I would recommend to use your spare OpAMP as impedance converter for the negative sense terminal and drive the reference ground.
Greetings from Germany.
Great vid! The "how to make a circuit that does XYZ" type videos are always appreciated by those of us with a software / digital electronics background!
-MCM
A compensation network is needed when putting a pass element like mosfet here. This is because it adds a phase lag. Of course you can just throw random caps to attenuate high frequencies and hope it will work well. But it does not guarantee stability. Only freq. response analysis over a range of temperature and loads (and supply voltage!) can ensure stability problem is solved. Also, gate resistance looks way bigger than it normally is. Most schemes use 10-100 Ohm gate resistors. Larger values may not actually solve problem, it hides it. This scheme is a perfect example of it: 12k was not enough, it needs a cap in the right place. Another thing is that electrolytic caps are bad in compensation loops as they have poor frequency response, bad tolerance and dry out.
could you do your version and post it
I love these "fly on the wall" videos. Learned heaps. Nice to see how you trouble shoot things. Thanks Dave!
This is awesome, I watch this blog all the time and just out of pure coincidence I was about to looking for a precision 1amp current source when I opened up my laptop earlier today!!
bravo, Dave I've always shied away from testing precision circuits on a plug-board like that. Thank you for showing that with just a little care, one can get meaningful results and some confidence that the properly-spun PCB WILL meet spec.
These intermediate level videos are the best :-)
So sad that you had to point out that Linear app note designs might not be tested to work. Jim Williams would be turning in his grave.
i dont understand wtf ur talking about half the time..but i enjoy listening to u
When I first started watching I didn't get anything but now I understand maybe... 20% or so. It's a strange way of learning, but hey.. it's free and enjoyable :P
+Yeti | Gameplay I agree. The funny thing is you only realize how much you know when you start talking to someone about it.
I would never try to do this kind of stuff on a breadboard. Mind the inductance of the wires. They are pretty much unpredictable.
Bloody good video Dave, lots of insightful info there for us juniors!!
niiiice!
Great video, great demonstrations! Each time I was screaming "well instead of talking about it, try it!" and you did, great!
The 12k gate resistor and the input capacitance of the mosfet creates a pole in the transfer function. The phase shift associated with the pole is causing the oscillation. The solution is to decrease the value of the gate resistor. This will move the pole much higher in frequency.
Yes. Something about 100-1000 Ohm could be fine.
I had the same oscilation problem with too large gate resistanse (>10k).
Great video. It's good to see that you registered that version of Dave CAD. I know he's particular about people using the pirated versions!
`Not even half a Bee`s dick...`I still laugh every time I view this.
the solderless breadboard you are using is "riddled" with inductance and stray capacitance. You need to hard-wire the circuit and keep the circuit tight.
What was the frequency range you were getting in the noise (couldn't see the scope time axis value) ? To start finding a value, I typically would then choose the RC time constant in something in that order of magnitude times some factor (for 470µF and 1.25Ω , RC is ~0.6ms).
that's what I would have done as well, would like it if he gets back to us about the switching residual instead of playing Russian roulette with those capacitors :)
Have a real similar circuit on the BenchBudEE for current output. I put a PTC resettable fuse on the high side V+
No trust in your current limiting circuit huh? :->
EEVblog if you short the V+ out in the field there isn't anything else to stop it from dump all the current the supply has on hand. Might be light for a battery powered thing like yours, mine is running off a filtered ATX power supply. Believe it can do 15A on the 12V rail? Yeah, $1 peace of mind! :-D
Chris Gammell And if in the field field with a lead acid :P That's some amps if you short stuff
Great Dave, I'm searching for a none "burning" solution for weeks, unfortunatly all those parts are currently out of stock ! Damn...
Slick resistors though. You did mention the retail price on those suckers, right? Don't get peoples' hope up! :-D
about $18 each from my quick googling
TheBdd4 $18 is nothing for a one-off precision current source like this. 2 terminal ones aren't much cheaper, if any.
They're $50 a pop on Digi-Key now and only a few are in stock. Not the same series either.
I'd like to build this but can't find a source for a 4 wire 1.25 ohm resistor. how much would I be messing this up if I just tapped sense off the pins of a 2 wire resistor? (ima noob)
would have been a good idea to measure that switching residual and work out the correct impedance for the damper. good work non the less.
One of those resistors costs TWENTY EIGHT DOLLARS.
At 25:00 Why shows the amp-meter 1.112A while the power supply's current limit kicks in at 1.100 A? Which is even visible on the PS.
Both devices were pretty close in value before and after this test. What's going on here?
The 2N2222 have an absolute rating of 800mA continuous collector current (based on farnell datasheet). You want to get 1A out of it That's probably why it oscillates.
Huh? I'm not using a 2N2222
EEVblog
I didn't watch part 1 of the video, thought you were just following the appnote.
And in no meaning You gave away for what You need that current source! ;-)
I guess 1 amp on four D-Cells: You're going to build a torch or an electrical foot warmer. :-)
Nice video. Thumbs up. Thanks
I don't understand why Dave didn't just use one op amp (low offset voltage) driving one N channel MOSFET. Non inverting input to the op amp connected to the precision Vref. The 4 terminal R on the source of the MOSFET to Gnd. Why use that feedback loop in the chip? Can't draw a circuit diagram here but I think there's a much simpler circuit that would do the same job.
That's exactly the design I am using.
Do you mean the original DC load design he used for testing?
Would increasing the load inductance help with stability ?
Is it safe to completely short the load terminals ?
How could you implement overvoltage protection at the load terminals (say someone sticks a charged capacitor on there)
How does it respond to high load resistance ? What is the min-max load resistance where the current will be in spec ?
Can it be damaged with ESD ?
So, when you changed the 10uf output cap to a 470uf cap, you matched the haromics of the suggested 4.7uf cap? Maybe you are just in the wrong octave O:-)
You started with a current source, you ended with a current sink!
How do you plan on handling the thermal issue with the FET? Just sink it?
Excellent vid!
If the goal is to avoid that 2mA sink into the sense out, why not just use the path from the bottom of the load to ground as output, rather than the top? Would that allow you to avoid the extra opamp?
Besides not giving you the added stability, of course.
12:41- “back from the future” lol
12:42 Back from the future!
"..not even half a bees dick.." LOL @19:37
I came here looking for all the promised teardown tuesdays. Instead I will watch more videos from Austria. :-p
Very nice video, keep them up!
Instead of adding the opamp, could you have instead moved the load to between the precision resistor and ground?
No, because then you'd have the chip current flowing through the load as well.
Thanks again for the great video, Dave! Loved all the explanations.
i always thought that the people who wrote the application circuits were wise overlords of the electronics world.
4:15 custom made to order? no wonder there's a minimum order
brilliant, thank you!
First scope trace looks like a line of armadillos
Hi. (I'm korean so my English is not good ^^;;)
I make a variable current source. It makes current ~1A.
But there's a problem that is too hot from the mosfet.
It's using 24V source for 0~20ohm Load.
How about you??
How much money I need to send to Vishay if I want them to send me 10 resistors like these 1R2500? :)
Ah Dave, you're such a tease. Chip ground and Sense ground both connected to you sense terminal.
Who can resist a resistor?
Someone come up with a current reference as a cloud service.
Very cool
Can somebody explain to me the concept of a four pin resistor?
+Thomas YT A 4 pin or 4 wire "Kelvin" connection is a more precise method of measuring to reduce error that could come into play due to resistances along the main current path. Search for "4 wire kelvin" on youtube. Afrotechmods and TechnologyCatalyst have very good tutorials on it to get you started
+Mike Novo Thanks, Mike. I think, I understand the principle. In the case of a 4 pin resistor, are two pins simply interconnected, so that I have one pair of contacts for the constant current source and one pair for measuring the voltage?
Exactly. The two inner pins are normally use to measure voltage and are trimmed for accuracy. If you don't have a 4 pin Shunt resistor, You want to solder your sense leads as close as possible to reduce error. I am very new at this and I'm sure others can chime in with more pearls of wisdom. Enjoy
Stop burning it, please!
19:02
Dave Cad license expired?
and renewed half way though the video....
is all about magic
Glad you found a work-around. However, what of the original design provided by Linear? I very much doubt they'd publish something that didn't work.
"Not even half a bee's dick" is a new one on me. We have another here in the US: "RCH" (red ---- hair). Quite vulgar, really. Among the crass, hormone-driven, insecure, semi-pubescent men in the electronics field encountered when I was growing up, this unit of measure came to frequent use.
Nice Video - but.
24;30 - Do you know what "good enough" means?
And please! Never ever ever do a Dave CAD with this black pen - it just don't look genuine ;)
Oh my god. Dave I like your videos, but well your circuit is again so bad. Anyway you connected at least pin 8 to ground now. Your circuit is for high side current source and you connect DMM on top of that ... and fixing that bug with an additional OPV is very bad. And there is more. You are feeding your complete GND current through the sense wire. You are wasting all those precision components by bad design.